Pandamonium

Cause panda and rucio don't work too good

This tells you if your jobs are done. And stuff like that.

No login required
Colors! (when you want them)
Gets input / output dataset names
Works with piping, other unix nice things

How I do dat?

Clone me
Add the directory to your PATH
Type pandamon [user.<your user name>]
Pretty colors!!!

You can add more of the task name if you want, and use wildcards (*). Wildcards are automatically appended to names that don't end in * or /.

Without any arguments the task name defaults to user.$RUCIO_ACCOUNT*.

Also try pandamon -h.

Other tricks

Get input/output dataset names

> pandamon -s IN <task name>
> pandamon -s OUT <task name>

Filter by user name

This is useful if you're running with group privileges. Set the environment variable GRID_USER_NAME to your full user name (the one that shows up on the top of the bigpanda page). Or specify one with --user.

Find input datasets for jobs in the `broken` state

You can do more useful stuff by piping through standard Unix utilities

> pandamon your.tasks > tasks.txt
> cat tasks.txt | awk '$1 ~ /broken/ {print $2}' | pandamon - -s IN

or (faster)

pandamon your.tasks -i broken -s IN

Filter by taskid range

Use to only display jobs in a specific range. This is useful for when you inevitably submit jobs with wrong parameters that you don't want to retry.

pandamon -r 12000-12100

Read the job user metadata

Now panda supports a userMetadata.json file for additional information in your job. Print it with

pandamon your.tasks -m

See this JIRA ticket where they plan to make it faster.

Testimonials

"I like colors" -- Chase Schimmin

"I found a bug" -- Danny Antrim (Fixed! Thanks Danny!)

"I tried to use it but it's python 3" -- also Chase

"I made a merge request. It was approved!" -- Alex

(I added python 2 support, and that's all that's currently supported)

I'll add other stuff too, if you want.

jburzy01/pandamonium